""" # --------------------------------------------------------------------------------- # -*- coding: utf-8 -*- ----------------------------------------------------------------------------------- # Copyright (c) Microsoft # Licensed under the MIT License. # Written by Bin Xiao (Bin.Xiao@microsoft.com) # Modified by Xingyi Zhou # Refer from: https://github.com/xingyizhou/CenterNet # Modifier: Nguyen Mau Dung (2020.08.09) # ------------------------------------------------------------------------------ """ from __future__ import absolute_import from __future__ import division from __future__ import print_function import os import torch import torch.nn as nn import torch.utils.model_zoo as model_zoo import torch.nn.functional as F BN_MOMENTUM = 0.1 model_urls = { 'resnet18': 'https://download.pytorch.org/models/resnet18-5c106cde.pth', 'resnet34': 'https://download.pytorch.org/models/resnet34-333f7ec4.pth', 'resnet50': 'https://download.pytorch.org/models/resnet50-19c8e357.pth', 'resnet101': 'https://download.pytorch.org/models/resnet101-5d3b4d8f.pth', 'resnet152': 'https://download.pytorch.org/models/resnet152-b121ed2d.pth', } def conv3x3(in_planes, out_planes, stride=1): """3x3 convolution with padding""" return nn.Conv2d(in_planes, out_planes, kernel_size=3, stride=stride, padding=1, bias=False) class BasicBlock(nn.Module): expansion = 1 def __init__(self, inplanes, planes, stride=1, downsample=None): super(BasicBlock, self).__init__() self.conv1 = conv3x3(inplanes, planes, stride) self.bn1 = nn.BatchNorm2d(planes, momentum=BN_MOMENTUM) self.relu = nn.ReLU(inplace=True) self.conv2 = conv3x3(planes, planes) self.bn2 = nn.BatchNorm2d(planes, momentum=BN_MOMENTUM) self.downsample = downsample self.stride = stride def forward(self, x): residual = x out = self.conv1(x) out = self.bn1(out) out = self.relu(out) out = self.conv2(out) out = self.bn2(out) if self.downsample is not None: residual = self.downsample(x) out += residual out = self.relu(out) return out class Bottleneck(nn.Module): expansion = 4 def __init__(self, inplanes, planes, stride=1, downsample=None): super(Bottleneck, self).__init__() self.conv1 = nn.Conv2d(inplanes, planes, kernel_size=1, bias=False) self.bn1 = nn.BatchNorm2d(planes, momentum=BN_MOMENTUM) self.conv2 = nn.Conv2d(planes, planes, kernel_size=3, stride=stride, padding=1, bias=False) self.bn2 = nn.BatchNorm2d(planes, momentum=BN_MOMENTUM) self.conv3 = nn.Conv2d(planes, planes * self.expansion, kernel_size=1, bias=False) self.bn3 = nn.BatchNorm2d(planes * self.expansion, momentum=BN_MOMENTUM) self.relu = nn.ReLU(inplace=True) self.downsample = downsample self.stride = stride def forward(self, x): residual = x out = self.conv1(x) out = self.bn1(out) out = self.relu(out) out = self.conv2(out) out = self.bn2(out) out = self.relu(out) out = self.conv3(out) out = self.bn3(out) if self.downsample is not None: residual = self.downsample(x) out += residual out = self.relu(out) return out class PoseResNet(nn.Module): def __init__(self, block, layers, heads, head_conv, **kwargs): self.inplanes = 64 self.deconv_with_bias = False self.heads = heads super(PoseResNet, self).__init__() self.conv1 = nn.Conv2d(3, 64, kernel_size=7, stride=2, padding=3, bias=False) self.bn1 = nn.BatchNorm2d(64, momentum=BN_MOMENTUM) self.relu = nn.ReLU(inplace=True) self.maxpool = nn.MaxPool2d(kernel_size=3, stride=2, padding=1) self.layer1 = self._make_layer(block, 64, layers[0]) self.layer2 = self._make_layer(block, 128, layers[1], stride=2) self.layer3 = self._make_layer(block, 256, layers[2], stride=2) self.layer4 = self._make_layer(block, 512, layers[3], stride=2) self.conv_up_level1 = nn.Conv2d(768, 256, kernel_size=1, stride=1, padding=0) self.conv_up_level2 = nn.Conv2d(384, 128, kernel_size=1, stride=1, padding=0) self.conv_up_level3 = nn.Conv2d(192, 64, kernel_size=1, stride=1, padding=0) fpn_channels = [256, 128, 64] for fpn_idx, fpn_c in enumerate(fpn_channels): for head in sorted(self.heads): num_output = self.heads[head] if head_conv > 0: fc = nn.Sequential( nn.Conv2d(fpn_c, head_conv, kernel_size=3, padding=1, bias=True), nn.ReLU(inplace=True), nn.Conv2d(head_conv, num_output, kernel_size=1, stride=1, padding=0)) else: fc = nn.Conv2d(in_channels=fpn_c, out_channels=num_output, kernel_size=1, stride=1, padding=0) self.__setattr__('fpn{}_{}'.format(fpn_idx, head), fc) def _make_layer(self, block, planes, blocks, stride=1): downsample = None if stride != 1 or self.inplanes != planes * block.expansion: downsample = nn.Sequential( nn.Conv2d(self.inplanes, planes * block.expansion, kernel_size=1, stride=stride, bias=False), nn.BatchNorm2d(planes * block.expansion, momentum=BN_MOMENTUM), ) layers = [] layers.append(block(self.inplanes, planes, stride, downsample)) self.inplanes = planes * block.expansion for i in range(1, blocks): layers.append(block(self.inplanes, planes)) return nn.Sequential(*layers) def forward(self, x): _, _, input_h, input_w = x.size() hm_h, hm_w = input_h // 4, input_w // 4 x = self.conv1(x) x = self.bn1(x) x = self.relu(x) x = self.maxpool(x) out_layer1 = self.layer1(x) out_layer2 = self.layer2(out_layer1) out_layer3 = self.layer3(out_layer2) out_layer4 = self.layer4(out_layer3) # up_level1: torch.Size([b, 512, 14, 14]) up_level1 = F.interpolate(out_layer4, scale_factor=2, mode='bilinear', align_corners=True) concat_level1 = torch.cat((up_level1, out_layer3), dim=1) # up_level2: torch.Size([b, 256, 28, 28]) up_level2 = F.interpolate(self.conv_up_level1(concat_level1), scale_factor=2, mode='bilinear', align_corners=True) concat_level2 = torch.cat((up_level2, out_layer2), dim=1) # up_level3: torch.Size([b, 128, 56, 56]), up_level3 = F.interpolate(self.conv_up_level2(concat_level2), scale_factor=2, mode='bilinear', align_corners=True) # up_level4: torch.Size([b, 64, 56, 56]) up_level4 = self.conv_up_level3(torch.cat((up_level3, out_layer1), dim=1)) ret = {} for head in self.heads: temp_outs = [] for fpn_idx, fdn_input in enumerate([up_level2, up_level3, up_level4]): fpn_out = self.__getattr__('fpn{}_{}'.format(fpn_idx, head))(fdn_input) _, _, fpn_out_h, fpn_out_w = fpn_out.size() # Make sure the added features having same size of heatmap output if (fpn_out_w != hm_w) or (fpn_out_h != hm_h): fpn_out = F.interpolate(fpn_out, size=(hm_h, hm_w)) temp_outs.append(fpn_out) # Take the softmax in the keypoint feature pyramid network final_out = self.apply_kfpn(temp_outs) ret[head] = final_out return ret def apply_kfpn(self, outs): outs = torch.cat([out.unsqueeze(-1) for out in outs], dim=-1) softmax_outs = F.softmax(outs, dim=-1) ret_outs = (outs * softmax_outs).sum(dim=-1) return ret_outs def init_weights(self, num_layers, pretrained=True): if pretrained: # TODO: Check initial weights for head later for fpn_idx in [0, 1, 2]: # 3 FPN layers for head in self.heads: final_layer = self.__getattr__('fpn{}_{}'.format(fpn_idx, head)) for i, m in enumerate(final_layer.modules()): if isinstance(m, nn.Conv2d): # nn.init.kaiming_normal_(m.weight, mode='fan_out', nonlinearity='relu') # print('=> init {}.weight as normal(0, 0.001)'.format(name)) # print('=> init {}.bias as 0'.format(name)) if m.weight.shape[0] == self.heads[head]: if 'hm' in head: nn.init.constant_(m.bias, -2.19) else: nn.init.normal_(m.weight, std=0.001) nn.init.constant_(m.bias, 0) # pretrained_state_dict = torch.load(pretrained) url = model_urls['resnet{}'.format(num_layers)] pretrained_state_dict = model_zoo.load_url(url) print('=> loading pretrained model {}'.format(url)) self.load_state_dict(pretrained_state_dict, strict=False) resnet_spec = {18: (BasicBlock, [2, 2, 2, 2]), 34: (BasicBlock, [3, 4, 6, 3]), 50: (Bottleneck, [3, 4, 6, 3]), 101: (Bottleneck, [3, 4, 23, 3]), 152: (Bottleneck, [3, 8, 36, 3])} def get_pose_net(num_layers, heads, head_conv, imagenet_pretrained): block_class, layers = resnet_spec[num_layers] model = PoseResNet(block_class, layers, heads, head_conv=head_conv) model.init_weights(num_layers, pretrained=imagenet_pretrained) return model